Linux RAID subsystem development
 help / color / mirror / Atom feed
* Timeout until degrade of RAID5 array
@ 2009-06-07 19:19 Gabriel Ambuehl
  2009-06-07 22:57 ` NeilBrown
  0 siblings, 1 reply; 2+ messages in thread
From: Gabriel Ambuehl @ 2009-06-07 19:19 UTC (permalink / raw)
  To: linux-raid

Dear List
Apologies if this is a FAQ but googling for an hour did not yield much of a 
result. 

My situation:
I have a NAS running 4 WD greenpower drives in RAID5. This works all nice and 
dandy, but since I do not actually need the NAS 90% of the time, I figured I 
would allow the drives to spin down to save on both power, noise and heat 
budget.

This works most of the times, but I've seen a few cases by now where one drive 
was kicked out of the array upon wake up. My assumption is the drive took a 
little too long to wake up and hence hit some form of a time out especially 
because I could easily rebuild the array with the same drive.

So Is there any way to increase the time that md would wait until removing a 
drive from an array?

Thanks for your time and kind regards,
Gabriel

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: Timeout until degrade of RAID5 array
  2009-06-07 19:19 Timeout until degrade of RAID5 array Gabriel Ambuehl
@ 2009-06-07 22:57 ` NeilBrown
  0 siblings, 0 replies; 2+ messages in thread
From: NeilBrown @ 2009-06-07 22:57 UTC (permalink / raw)
  To: Gabriel Ambuehl; +Cc: linux-raid

On Mon, June 8, 2009 5:19 am, Gabriel Ambuehl wrote:
> Dear List
> Apologies if this is a FAQ but googling for an hour did not yield much of
> a
> result.
>
> My situation:
> I have a NAS running 4 WD greenpower drives in RAID5. This works all nice
> and
> dandy, but since I do not actually need the NAS 90% of the time, I figured
> I
> would allow the drives to spin down to save on both power, noise and heat
> budget.
>
> This works most of the times, but I've seen a few cases by now where one
> drive
> was kicked out of the array upon wake up. My assumption is the drive took
> a
> little too long to wake up and hence hit some form of a time out
> especially
> because I could easily rebuild the array with the same drive.
>
> So Is there any way to increase the time that md would wait until removing
> a
> drive from an array?

No - this is not a function of md at all.

You should be asking "is there some way to get the SCSI/SATA/whatever
controller to wait a bit longer before failing an IO request?"
And that questions should, of course, be directed to developers
of the driver for whatever sort of disk controller you have.

Good luck.

NeilBrown



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2009-06-07 22:57 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-06-07 19:19 Timeout until degrade of RAID5 array Gabriel Ambuehl
2009-06-07 22:57 ` NeilBrown

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox